Black Forest Labs released FLUX.3 Image Klein, with day-one support in Diffusers. The model supports text-to-image and multi-reference image-to-image, and adds layout control by specifying bounding boxes in prompts. It ships two open checkpoints: a distilled 16-step no-CFG version and a base 50-step CFG version. The distilled checkpoint generates a 1024×1024 image in about two seconds on an H100, with NVFP4 and FP8 quantized weights and LoRA training scripts provided.
Black Forest Labs has released its new flagship image model, FLUX 3 Image, now available on OpenRouter. The model supports pixel-level multi-turn editing, bounding box layout, up to 4K generation, and up to 10 reference images, with commercial weights for enterprise and open weights coming in the following weeks.
Black Forest Labs' FLUX Video Edit model has gone live on OpenRouter, supporting text instructions to remove or replace objects and characters in video, change settings, restyle footage, and translate dialogue with lip sync, without masks or per-frame controls.